https://bugzilla.novell.com/show_bug.cgi?id=657360
https://bugzilla.novell.com/show_bug.cgi?id=657360#c4
Petr Uzel
I cannot see a reason why informing the kernel should fail. A few seconds earlier setting lvm and raid off works fine. Maybe the usual suspect udev?
Yes, that would be my guess: udev or whatever called by udev. In parted-2.3, there is already the logic "inform the kernel; if (!success) sleep and retry". The maximum sleep is one second [*]. I really have no idea (nobody seems to have) how to make informing the kernel about PT changes 100% reliable without rebooting. Two ideas how to make it at least more reliable than it is now: 1/ increase the max sleep time [*] to - let's say - 2 seconds to give udev (or whatever it is) more time to scan the disk (which is as far as I understand triggered by the uevent that is generated when the partition is added) and close it. 2/ as far as I remember, we have usually seen these "can not inform the kernel.." messages after setting some partition flag. Furthermore, as far as I can see, the kernel does not care about the partition flags, so I think I could patch parted not to inform the kernel about any changes after 'set flag state' command. Arvin, does this make sense? Or do you have any better idea? Thanks. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.